Robert T. H. Chi

Distributed intelligent executive information systems

Abstract Executive information systems (EIS) have been successfully implemented in many organizations. Of all the various EIS commercial products, only one (Executive Edge) presents limited artificial intelligence (AI) capabilities. Yet, the ability to include various problem solving agents for collaboratively information processing, filtering and presentation, is highly desirable. It is possible that the successful EIS systems of the future will be built around AI components (expert systems, learning mechanisms and so on..), so that more efficient and effective information processing for executives can be achieved. Since much of executive processing involves complicated problem domains, a singles AI agent effort may be insufficient when the information is broad in scope and complicated in nature. For such situations we propose in this paper a framework called distributed intelligent executive information system (DIEIS). This framework illustrates how multiple resources (consisting of knowledge learning, reasoning, filtering and presentation) can be combined for information processing in an EIS environment. For example, a particular piece of information may be refined and presented based on past experiences and current practices in a particular problem domain with the help of both an expert system and neural computing. The DIEIS framework allows multiple agents to work collaboratively to help complex information processing.

Generalized case-based reasoning system for portfolio management

Abstract A case-based reasoning system (CBRS) is appropriate for an experince-rich domain, while a rule-based system performs reasonably well in a knowledge-rich application environment. Performance of a CBRS suffers when past experience is not readily available. A generalized case-based reasoning system (GCBRS) is proposed to remedy this weakness by incorporating domain theories represented as generalization rules. With these rules, previous experience (stored as cases) can be generalized so that the possibility of solving a new case is higher than it would be when case-based reasoning is used alone. The architecture and the inference mechanism of a GCBRS are discussed in this article. A portfolio management support system based upon the proposed GCBRS architecture is presented to demonstrate the feasibility of using GCBRS for developing a decision support system in a knowledge-poor and experience-poor domain. This article concludes with a discussion of future research.

Reasoning by coordination: an integration of case-based and rule-based reasoning systems

A case-based reasoning system (CBRS) is appropriate for an experience-rich domain, while a rule-based system performs reasonably well in a knowledge-rich application environment. A CBRS mainly uses past experiences as problem-solving tools, and therefore its capability can be limited when previous experiences are not a good representation of the whole population. On the other hand, a rule-based reasoning system needs a well constructed domain theory as its reasoning basis, and it does not make use of the knowledge embedded in past experiences. In the paper, a multiagent cooperative reasoning system (MCRS), which integrates an inductive-reasoning agent (case-based reasoning system) and a deductive-reasoning agent (rule-based reasoning system), is proposed to solve problems through the cooperation of both agents. The architecture and the inference mechanism of an MCRS are discussed in this paper. A personnel-evaluation support system, based on the proposed MCRS architecture and implemented in prolog, is presented to demonstrate the feasibility of using the MCRS to solve problems in a knowledge-poor and experience-poor domain.

DKAS: a distributed knowledge acquisition system in a DSS

Knowledge acquisition is the process of accumulating new information and relating it to what is already known. Knowledge acquisition has been regarded as the bottleneck in knowledge-based systems development. In this paper, a distributed knowledge acquisition system (DKAS) is introduced for automating decision rules construction from a set of examples in a decision support system. DKAS has the potential to include various learning mechanisms and employs a multi-agent and parallel processing paradigm. The implementation of a DKAS integrates inductive and deductive learning methods that use different learning strategies. A stock selection problem is used to demonstrate the effectiveness of DKAS in solving classification type problems. The performance of the DKAS in portfolio management is compared to the performance of the NYSE and the S&P 500. The results indicate that the rules derived from using the DKAS outperform both the NYSE and the S&P 500.

The application of SOM as a decision support tool to identify AACSB peer schools

For a business school, the selection of its peer schools is an important component of its International Association for Management Education (AACSB) (re)accreditation process. A school typically compares itself with other institutions having similar structural and identity-based attributes. The identification of peer schools is critical and can have a significant impact on a business schools accreditation efforts. For many schools the selection of comparable peer schools is a judgmental process. This study offers an alternative means for selection; a quantitative technique called Kohonens Self-Organizing Map (SOM) network for clustering. In this research, we first demonstrate the capability of SOM as a clustering tool to visually uncover the relationships among AACSB-accredited schools. The results suggest that SOM is an effective and robust clustering method. Then, we compare the results of SOM with that of other clustering methods, such as K-means, Factor/K-means analysis, and kth nearest neighbor procedure. The objective of this study is to demonstrate that a two-dimensional SOM map can be used to integrate the results of various clustering methods and, thus, act as a visual decision support tool.

Understand user preference of online shoppers

The tremendous growth of the Internet has created opportunities for consumers and firms to participate in an online global marketplace. It is conceivable that in the future every person with access to a computer will interact with firms marketing on the Internet. We foresee that advances in electronic commerce dramatically alter the structure of businesses, especially in the marketing area. We extend the literature on marketing channel functions to include the Internet as a new option for selling products/services directly to customers. Important factors that inference the behaviors of online shoppers are identified. A classification scheme is used to categorize products/services selling on the Internet based on characteristics such as tangibility and price from the buyers perspective. The classification scheme helps management to understand the difference in user preference over different product categories. A survey of experienced online shoppers was conducted to validate the scheme.

The Effect of Sample Size on the Extended Self-Organizing Map Network for Market Segmentation

Kohonens Self-Organizing Map (SOM) network maps input data to a lower dimensional output map. The extended SOM network further groups the nodes on the output map into a user specified number of clusters. Kiang, Hu and Fisher used the extended SOM network for market segmentation and showed that the extended SOM provides better results than the statistical approach that reduces the dimensionality of the problem via factor analysis and then forms segments with cluster analysis. In this study we examine the effect of sample size on the extended SOM compared to that on the factor/cluster approach. Comparisons will be made using the correct classification rates between the two approaches at various sample sizes. Unlike statistical models, neural networks are not dependent on statistical assumptions. Thus we expect the results for neural network models to be stable across sample sizes but may be sensitive to initial weights and model specifications.

Understand Corporate Rationales for Engaging in Reverse Stock Splits - A Data Mining Application

There has been much written on the individual topics of bankruptcy prediction, corporate performance, and reverse stock splits. However, there is little research into the relationship between reverse stock splits and corporate performance as well as bankruptcies. The purpose of this study is to provide and empirically support rationales for reverse splits by classifying reverse splitting firms into two groups, those declaring bankruptcy within 2 years and those remaining solvent. The apparent rationales for engaging in reverse splits differ between the two groups, i.e., weak firms attempting to increase their stock price while solid firms seeking to reposition their stock in the market. Two alternative approaches, Altmans Z-scores and artificial neural networks, are used for classifying reverse splitting firms into the two groups. A comparison is then made of the relative success of Z-scores and neural networks in the classification. This study should generate an understanding of corporate rationale for engaging in reverse splits and the relative success of Z-scores and artificial neural networks in forecasting the two groups.

Using an integrated model learning system to construct the model base of a decision support system

The Model Base Management System (MBMS) in a Decision Support System (DSS) has become increasingly important in handling complicated decision problems. Traditionally, models are retrieved from the human experts. This approach has the disadvantages of low productivity and being subjective by different individuals. Although recent research projects include how to apply Artificial Intelligence approaches to automatic model retrieval, none of them provides an efficient and effective way to produce large‐scale complicated models. In this article, an Integrated Model Learning System is introduced to learn new models from model instances. This system basically uses the Multiple Instance Explanation Based Generalization (miEBG) approach which is a modification of the Multiple Explanation Based Generalization (mEBG) approach to explain and generalize model instances by the Domain Theory. Different from other machine learning approaches, this approach takes advantages of using the Domain Theory to learn multiple explanation trees from instances, and then combines all trees to form a complete explanation tree which will be generalized into a useful model. In general, this approach provides the Model Base Management System an automatic way to retrieve new and more complete models from related instances in a self‐learning fashion.

Service oriented automated negotiation system architecture

Automated negotiation has become the core of the intelligent e-commerce. Traditional research in automated negotiation is focused on theory about negotiation protocol and strategy. However, the application of automated negotiation system has lagged far behind. This paper discusses the reason for such a situation, and proposes a technology roadmap for the development of automated negotiation system using the software agent technology. In order to find a way to the applicable application of the system, this paper proposes an application architecture based on SOA and web services technology.